Jan Tattermusch
0cb7fef406
Revert "Split bm_fullstack into pieces, reuse infrastructure across other microbenchmarks"
8 years ago
Craig Tiller
12352b2c5c
Add test of server builder, including channel args
8 years ago
Craig Tiller
4751c28b0b
Add channel arguments to server builder
8 years ago
Craig Tiller
e067b6e119
Resolve sanity failures
8 years ago
Craig Tiller
f09ec59222
Fix mac build
8 years ago
Craig Tiller
1128d462c3
Pollset benchmark
8 years ago
Craig Tiller
52bd441b25
Get back trickle
8 years ago
Craig Tiller
12d22e472b
Further splitting
8 years ago
Craig Tiller
62ac65b183
Split bm_fullstack unary ping pong out
8 years ago
Craig Tiller
627a59892a
Prep for splitting bm_fullstack
8 years ago
Craig Tiller
5e3215338f
Move helpers to a common place, use them everywhere
8 years ago
Craig Tiller
0a3d5f90a9
Revert "Enable lto for performance testing"
8 years ago
Muxi Yan
5c2d7e287b
Boost grpc version to v1.1.4
8 years ago
Craig Tiller
f9abb2921b
Reserve opt for external builds, use lto for profiling
8 years ago
Craig Tiller
295df6da9a
Add a slice type that shares a refcount with a transport stream
8 years ago
Craig Tiller
5e9bdb784d
Enable lto
8 years ago
Craig Tiller
edbf2b9d13
Add a spinlock type.
...
Useful for situations where we need to repeatedly trylock, not useful
for cases where we need to lock (due to spinning).
Add a variant of sync_test to test it (with the same tests we run for
gpr_mu).
Add a benchmark to bm_closure to demonstrate single threaded performance.
8 years ago
ncteisen
40e1e99f62
Fix load_reporting linker error
8 years ago
Ken Payson
fd6cc7de7b
Bump version to v1.1.3
8 years ago
ncteisen
3da3ce3c29
Add error test
8 years ago
Mark D. Roth
d6d192d005
Retry throttling implementation.
8 years ago
Craig Tiller
523d54beeb
Refine to just be a HPACK benchmark
8 years ago
Yuchen Zeng
a9d8a157be
Add parse_address_test
8 years ago
Craig Tiller
2f792d9d16
Initial http2 benchmarks
8 years ago
murgatroid99
acd0ba0ca3
Fix segfault in Node server destructor
8 years ago
Craig Tiller
122e4500d1
Exclude poll, poll-cv for bm_fullstack (its not interesting)
8 years ago
David Garcia Quintas
e854357117
Reintroduced golden_file_test
8 years ago
Craig Tiller
f3dec9c995
Add counters for rmw atomic operations to microbenchmarks
8 years ago
David Garcia Quintas
1f980cf7bb
Removed useless cpp/codegen/golden_file_test
8 years ago
Craig Tiller
19f3ea2c20
Start bringing microbenchmarks to Bazel
8 years ago
Craig Tiller
8506219a15
Remove internal api canaries: they have never solved anything for us
8 years ago
murgatroid99
1191b7202d
Improve Node and libuv testing and test coverage
...
Allow Node tests to run with or without UV, change default version to 7, add some portability tests. Also make some more core tests work with libuv
8 years ago
Craig Tiller
cba864bfbe
Eliminate port server env var, force its use always, consolidate code
8 years ago
David Garcia Quintas
448191cb9a
Temporarily disable lb policies test
8 years ago
Craig Tiller
b8e2bca4eb
Add a benchmark for metadata primitives
8 years ago
Craig Tiller
3f06e3f8a6
Increase timeout for bm_fullstack
8 years ago
Craig Tiller
264307fa3d
Add benchmarks for grpc_error
8 years ago
Craig Tiller
fd044b8b22
Call creation benchmark
8 years ago
Craig Tiller
14ae5381af
Add microbenchmark of single-threaded CQ operations
8 years ago
Craig Tiller
fe5f497f77
Add a test that measures flow control stalls in a deterministic way
8 years ago
David Garcia Quintas
f31f096274
Use call context to propagare LR costs
8 years ago
Craig Tiller
b3f34b6c2b
Add benchmarks of closures, combiners, exec_ctx primitives
8 years ago
Mark D. Roth
1dcd922ce6
Remove initial_connect_string hack.
8 years ago
David Garcia Quintas
012915045f
Secure naming support for gRPCLB
8 years ago
Harvey Tuch
5f3cfe960f
Fix read from uninitialized memory bug in GrpcBufferWriter.
...
This commit fixes an issue in which the following sequence of operations
leads to use of uninitialized memory:
1. Caller invokes GrpcBufferWriter::Next(), and then makes use of 8191
bytes in the returned buffer (which is 8192 bytes in size).
2. Caller then returns the unused single byte via
GrpcBufferWriter::BackUp(). This method invokes
g_core_codegen_interface->grpc_slice_split_tail(), which causes
backup_slice_ to be a grpc_slice with one byte.
3. At the next invocation of GrpcBufferWriter::Next(), a reference to
the single byte grpc_slice is returned to the caller.
The problem here is that the returned reference is to the inlined buffer
in the grpc_slice, which is resident in slice_, not the location of the
buffer inside slice_buffer_ after
g_core_codegen_interface->grpc_slice_buffer_add() in
GrpcBufferWriter::Next(). As a result, any data the caller writes to the
returned void* data is lost.
The solution is to avoid inlined backup slices.
8 years ago
Craig Tiller
b038beb724
Add counters for mutex acquisitions, expose in bm_fullstack
8 years ago
Piotr Pietraszkiewicz
cc08a4f24f
Fix issue #9602
...
This is achieved by putting `grpc++_codegegen_proto` on the
`filegroup` of `grpc++`.
8 years ago
murgatroid99
2b3e12ceee
Update version to 1.1.2
8 years ago
Craig Tiller
1bb53718bf
Disable writes_per_rpc on Windows
8 years ago
Craig Tiller
a4354b0185
Speed up bm_fullstack test
8 years ago